Los cuidadanos como detectores de la calidad del aire. Leonor Tarrasón, Directora de soluciones sostenibles

Transcription

1 Los cuidadanos como detectores de la calidad del aire Leonor Tarrasón, Directora de soluciones sostenibles

2 Traditional Air Quality Monitoring Is there another way? Large Complex High-maintenance Expensive Very sparse Bidgee

3 Satellite-based air quality monitoring Courtesy NASA

4 There might be

5 AIR QUALITY PARTICIPATORY SERVICES

6 Sensing the city with static nodes Information at citizen level

7 Sensing the city with buses We measured at the source - with buses

8 Sensing the city with bicycles We measured where people cycle

9 Sensing the city with people We measured where the people walked NO 2 +O 3 AQ Temp UV

10 Opportunities Supplement routine ambient air monitoring networks Monitor personal exposure Monitoring at the source and where people is affected Stimulate citizen participation and increase awareness

11 ECLECTIC Big Data Services

12 MISSION Reduce the emissions in the city. Improve the cabin air

13 CAMS Air quality models Sensors in cars and Trucks Air quality station Raw sensor data Big Data Services Smart city services Green zones Detailed air quality monitoring Car services Sensor calibration Air quality information Green route planner Smart air inlet

14 Smart air inlet Combining model and sensor -data to predict air quality enables the car to have a smart control of the air inlet to the cabin. Reduce the air inlet when the air is polluted. Maximise the air inlet when the air is clean.

15 Green zones Establish green zones at location where we want low air pollution; Schools, kindergartens, hospitals Automatic establish green zones where the emissions are high. Switch hybrid cars into electric when they enter green zones. Switch fuel driven cars into eco mode, ref. VW dieselgate

16 CAN WE USE LOW-COST SENSORS FOR AIR QUALITY MANAGEMENT? Big Data Services

17 Research projects EU: FP7 CITI-SENSE EMIIA II CITI-SENSE-MOB FP6 MEMORI COST EUNetAir Norwegian Research Council: INNOSENSE ECLECTIC CrowdAir EU networks: AQUILA CEN FAIRMODE

18 Air Patrol and Flow by Plume labs No information about the performance

19 Airbeam No information about the performance in their website but independent field calibration by AQMD

20 Low-cost sensor platforms are available to the public but information about performance is lacking

21 The AQMesh platform v3.5 Information extracted from AQMesh documentation in CITI-SENSE project Environmental Instruments Ltd, UK, Can we reproduce these values?

22 Laboratory evaluation: set-up CEN Gas analyzers Thermostatic bath Low-cost nodes Gas Sensor type CO NO 2 NO O 3 Electrochemical CO-B4 Electrochemical NO2-B42F Electrochemical NO-B4 Electrochemical OX-B421 Field evaluation: set-up Gas Analyzer CO Teledyne API 300E (EN14626) NO x Teledyne API 200A (EN 14211) O 3 Teledyne API 400 (EN 14625) Gas CO NO 2 NO O 3 Sensor type Electrochemical CO-B4 Electrochemical NO2-B42F Electrochemical NO-B4 Electrochemical OX-B421 Gas Analyzer CO EC Serinus 30 (EN14626) NO x EC Serinus 40 (EN 14211)

23 Field evaluation results: calibration Low accuracy Low precision Good accuracy Good precision Low accuracy Low precision CO NO NO 2 O 3 Low accuracy Good precision A good performance in the laboratory is not indicative of a good performance in field. Correlations significantly lower in the field than in the laboratory. Necessary to calibrate the sensors in the field. The results show that even for identical sensors and platform, the performance can vary sensor to sensor. Challenge in ensuring sensor measurement repeatability.

24 Field evaluation results: long-term performance NO, node Clear change in the behaviour during the 6 months co-location period due to varying weather conditions and atmospheric concentrations. The variation in the calibration parameters month to month can be significant. This can lead to increased errors and biases that can pass unnoticed once the nodes are deployed in the field.

25 Field evaluation results: dependence on meteorological conditions The response of each sensor to weather conditions is unique, and it is necessary to evaluate each sensor individually. We can have false increases in concentrations due to changes in temperature.

26 Field evaluation results: dependence on the location Node CO NO NO 2 O 3 PM 10 PM 2.5 Coef. determination (r 2 ) Lab Coef. determination (r 2 ) Field (dense traffic) Coef. determination (r 2 ) Field (calm traffic) Slope Lab Slope Field (dense traffic) Slope Field (calm traffic) Intercept Lab Intercept Field (dense traffic) Intercept Field (calm traffic) The linear calibration parameters are different when the node is located in a traffic-saturated environment or at a traffic-calm environment. It is important to calibrate the nodes in an environment similar to the one in which they would be deployed (or better, to perform in-situ calibration at the deployment site).

27 The CleanAIR project: AirSensEUR A multi-sensor platform designed for the monitoring of air pollution at low concentration levels. Completely open platform (both hardware and software) Designed/developed by JRC in Ispra Three AirSensEUR units tested as part of the CleanAir project with Telia and Oslo Kommune Calibration methodology developed inhouse at NILU (multiple linear regression and neural networks tested) Field calibration for NO 2 carried out at Kirkeveien station AirSensEUR unit with Teflon enclosure Co-location at Kirkeveien station

28 First results for NO 2 look promising

29 First results for NO 2 look promising Three AirSensEUR units (P1, P2, and P3) have been co-located at Kirkeveien during February and March R 2 values are consistently around 0.8 and higher.

30 Key messages A good performance in the laboratory is not indicative of a good performance under realworld conditions. Necessary to perform field calibration for each sensor node individually. Performance and field calibration parameters vary spatially and temporally, as they depend of the meteorological conditions and the atmospheric composition. We can not ensure absolute values (e.g. the concentrations are lower or higher than the limit value), but for some pollutants and nodes we can get coarse information (e.g. the air pollution is lower or higher than yesterday). Field calibration still represents a challenge. Necessary to employ more sophisticated techniques than linear calibration. After data processing we can extract useful information and generate detailed air quality maps. Soon ready for Copernicus?

31 Núria Castell Philipp Schneider Franck René Dauge THANK YOU